A: For now, you will have to find working parameters manually. Note though, that there are many possible parameters to make a network behave like in reality. Usually a set of different parameter can change the magnitude of some expression profiles, while their overal shape can stay the same. So it is usually still feasible to find a set of appropriate threshold and rate values.

If it just doesn't seem possible to find working values, then perhaps you should start hypothesizing if something should be changed in the supposed network structure. How long you have to seek for parameters before you can start doubting on your network is of course difficult to tell. But when your network has a large number of components, automatized searching can be helpful.

That's why we are currently working on parameter estimation for SIM-plex, against (roughly) known target profiles. We are currently working on this feature and plan to make it available in a next version.
 

A: You can define two different components: one for the mRNA and one for the protein associated with the gene. You should do that when transcription and translation show a clear difference in behaviour.
But with the current status of research, for many genes there is often only mRNA level knowledge available (microarrays), or only protein knowledge (Western Blot), and people have to work with the assumption that mRNA upregulation will correspond with protein level rise, though a bit slower.
Therefore SIM-plex makes it possible to define transcriptional creation and transcriptional blocks, as well as chemical creation and chemical break-down; and these four actions can be combined together for each geneproduct. See the "Note about creation, block and break-down" in the tutorial or in the syntax-part of the manual.
 

A: This is not what SIM-plex is made for.  SIM-plex is made for modeling regulatory interactions. SIM-plex's underlying mathematic model is (for now) only based on stepwise activations, because this type of mathematics makes it possible to translate if-then statements so transparently into differential equations.

(The non-regulatory if-then-transform statements are in fact already an extension that is quantitatively less correct in the used mathematical framework.  But it is still a very usable extension when you model in the context of (switch-like) genetic regulatory networks in which temporary disactivations of proteins by phosphorylation etc. are crucial, intermediary steps in the network.)

Also note that for more gradual binding and activation, you could define an activation with two or three steps, based on two or three consecutive thresholds for the activating component.
 

A: SIM-plex can export the results to a (tab-delimited) textfile readable by Excel. So while the plot windows show results quickly, creating fancy graphs can happen with a specialized program, outside the simulator.